A Tool for Analyzing Structure and Function Parameters of Artificial Neural Networks

Traditional analysis of artificial neural networks (ANNs) is mostly restricted to "visual inspection" and measuring the performance of the trained network. However, for a more detailed analysis structural parameters (e.g., number of connections, connectivity) and functional parameters (e.g., neuron activities under different network inputs) would be of great interest. netGlass should enable the comparison of (evolved) ANNs and identify potential similarities in different networks possibly using measures borrowed from the analysis of biological neural networks (BNNs). The tool will be employed for experiments in the field of computational neuroethology (carried out in cooperation with biologists).

Project Student

Required Knowledge

Acquired Knowledge


Financial Support

Project Duration